Split PDF

Enter a maximum size in MB. Starting from page 1, each file takes as many consecutive pages as fit: candidate files are actually built and their size measured, so the number is the downloaded file's real size, not an estimate. Pages are never reordered. A page that is over the limit by itself cannot be made smaller by splitting, so it becomes a file of its own above the limit, and the finished panel names that page; Compress PDF is the tool that can shrink it.

Yes, when its chapters are bookmarked: choose By bookmarks, and a new file starts at the page each top-level bookmark opens; nested bookmarks do not cut, and neither does a bookmark to a web address. Pages before the first bookmark become a file of their own, and two bookmarks on one page start one file. The preview shows which bookmark begins each file. If the PDF has no bookmarks that open a page, the option is greyed out with that reason.

Yes: choose By text and type the word or phrase. Every page whose text contains it starts a new file, and pages before the first one become a file of their own; the preview names the page that starts each file. Spaces and line breaks are compared loosely, so a phrase that wraps onto a second line still matches, and capitals are ignored unless Match case is ticked. It reads the PDF's own text, so a scanned page with no text layer cannot start a file — it is named under the field and stays with the pages before it. OCR PDF adds a text layer first.

Separate files with commas. A single number such as 8 is a one-page file, 2-5 includes both ends, 10- runs to the last page and -3 starts from page 1. An entry that cannot work stops the preview with a message: a page past the end is named along with the PDF's page count, 5-3 is flagged as backwards, and 0 or a word is refused. A page you leave out of every range is in none of the outputs. To keep most of a document and drop a few pages, Delete Pages does that in one file; for a hand-picked set of pages combined into one new PDF, Extract Pages is the closer fit.

Each piece takes the original name, a sequence number and its page span, for example report-01-pages-1-3.pdf or report-12-page-40.pdf, in every mode. With ten or more pieces the number is zero-padded so they sort in order, and the ZIP that holds multiple pieces is named after the original with -split added. Ranges merged into one file download as that one PDF, named after the original with -split added.

Not in this tool. Unlock PDF removes the encryption when you type a password the file accepts; its password field cannot be left empty, so a file that only restricts printing or copying needs its owner password there. Unlock PDF is not in-browser: the PDF and the password go over TLS to pdf.xyz's secure server, the input is deleted when the job finishes, the download link expires after 60 minutes, and the stored copy is erased by a daily storage rule after that. The split you do afterwards stays in your browser.

Not fully. The page and its processing engine load over the network, and parts of the engine are fetched the first time a step needs them. The processing itself runs on your device. Chinese, Japanese or Korean text whose font is not embedded in the PDF needs a character map, downloaded from our site the first time it is needed, and that step needs a connection.

Up to 100 MB per file.

After your file, then each piece's number and the pages it holds — for example report-1-pages-1-3.pdf and report-2-page-4.pdf. More than one piece downloads as a ZIP; a single piece saves as a PDF.

Yes. Bookmarks and form fields travel with the pages you keep: a bookmark to a kept page still opens it, and the kept pages' fields stay fillable with their values. A bookmark that opens only a page you left out is not carried. The file also keeps the title and author, page labels and the document language. Attached files and a tagged PDF's accessibility tags are not carried.